Comparison of accidental pediatric scald burns in a tertiary care center: hot cauldron burns versus accidental spill burns

Background: Scald burns result from exposure to high-temperature fluids and are more common in the pediatric age group. They occur mainly by two mechanisms: (i) spill and (ii) immersion (hot cauldron) burns. These two patterns differ in clinical characteristics and outcomes. Scalds cause significant morbidity and mortality in children. The objective of this study was to compare accidental spill burns and hot cauldron burns in a hospital setting. Methods: An analytical cross-sectional study was conducted by reviewing the secondary data of scald cases admitted during the years 2019 and 2020 in a burn-dedicated tertiary care center. Total population sampling was adopted. Data analysis was done partly using SPSS, version-23, and Stata-15. Mann Whitney U-test and Chi-square/Fisher's exact test were done appropriately to find associations between different variables. Binary regression analysis was performed taking mortality events as the outcome of interest. Results: Out of 108 scald cases, 43 (39.8%) had hot cauldron burns and 65 (60.2%) had accidental spill burns. Overall mortality was 16 (14.8%), out of which hot cauldron burns and accidental spill burns comprised 12 (75.0%) and 4 (25.0%), respectively. Binary logistic regression analysis showed the type of scald, age, and Baux score found to be associated with mortality. Every one-year increment in age had a 29% lower odds of occurrence of mortality event (adjusted odds ratio [OR], 0.71; 95% confidence interval [CI], 0.50-0.99, p=0.042). Likewise, every one-point increment in Baux score was associated with 19% higher odds of mortality (adjusted OR, 1.190; 95% CI, 1.08-1.32; p<0.001). Conclusions: Accidental spill burn was more common but mortality was significantly higher for hot cauldron burns. The risk of mortality was significantly higher in burn events occurring outside the house, and burns involving back, buttocks, perineum, and lower extremities.


Introduction
Scald burns are the most common types of burn injuries in children throughout the world. 1,2 These injuries are the leading causes of morbidity and mortality in the pediatric population. 3,4 They have long-term physical, psychological and economic impacts on the patients and their families. The majority of scald injuries occur at home and most of them are accidental and preventable. [5][6][7] Scalds result from exposure to hot liquids or steam. They occur mainly by two mechanisms: (i) spill and (ii) immersion burns. 8-10 Accidental spill burns are due to spillage of hot liquid such as boiling water, tea, milk, oil, soup, etc. These burns usually happen if a child gets in the way of an adult carrying hot fluids or accidentally play with the utensils filled with hot liquid and spill over their body. Immersion burns take place when children fall or put their hands/feet into a vessel containing hot liquid such as bathtub, tea/coffee pots, hot cauldrons, etc. 3,6,[11][12][13][14] Past studies have shown that the patterns of these two scalds differ widely for different age groups, gender, and body parts involved, with differences in hospital stay and outcomes secondary to extent of burns. 6,[15][16][17] The study of different types of scalds can help to control or reduce the predisposing factors and ultimately help in formulating plans and strategies for prevention. This study aimed to compare accidental spill and immersion (hot cauldron) scald burns in different aspects in a tertiary care center. Moreover, the factors affecting mortality were also studied.

Study setting
The study was conducted in the tertiary care center Kirtipur Hospital, Kathmandu, Nepal. It is a national referral center for the management of burn injuries. This hospital has 100 beds in total. Out of these, 32 beds are allocated for burn cases only; of which eight are in Burn Intensive Care Unit (BICU). It takes care of the whole spectrum of burn cases from acute care to the full range of reconstructive services and rehabilitation.

Study description and variables
This was a single-center, analytical, cross-sectional study carried out by reviewing the secondary data of scald burn cases admitted during two years (2019 and 2020). In this study, comparison was done at two different levels. Firstly, the types of burns, namely hot cauldron burns (HCB) and accidental spill burns (ASB) were analyzed in terms of socio-demographic factors, clinical features at presentation, in-hospital treatment measures and associated outcome (i.e. mortality). Next, comparison was done between survivors and non-survivors in terms of socio-demographic factors, presenting features and treatment measures. Out of all scalds, only pediatric cases (age < 18 years) were included in our study. Secondary data was collected from hospital records maintained in electronic form which included all of the acute scald cases admitted to the Plastic Surgery Ward and BICU. Scald cases managed on an outpatient basis were excluded from the study. Total population sampling was used. Those cases with incomplete records were excluded.
The data extracted from the patients' electronic records and stored in our database included demographic information like age, gender, and address of residence. All the immersion injury cases presented to our center had occurred due to falling into cauldrons filled with boiling water containing shredded straw and husk to feed cattle, which is commonly used in Nepal. Spill burns occurred due to spillage of hot liquids (boiling water, tea, coffee, milk, and oil). The mechanism of scalds was categorized into two broad groups: hot cauldron burns (HCB) and accidental spill burns (ASB). Data related to burn injury included type (mechanisms) of scalds, place of burn, pre-hospital intravenous fluid use, presence of infection at presentation, total body surface area, and body parts involved. Likewise, data of in-hospital interventions such as blood transfusion, escharotomy, necrosectomy, tangential excision, second excision, use of graft and/or flap, amputation were also taken. Outcome variables like duration of hospital stay and mortality were also noted.

REVISED Amendments from Version 1
This is the first revised version of the article after post-publication reviewers' comments. In this version, we have clarified the objectives of the study. The conclusion section has been edited to make it more aligned with the objectives. Minor grammatical errors have been corrected. Previously, we had used the term "multinomial logistic regression" which was a blunder. We have replaced it with binomial logistic regression because our dependent variables at two different levels of comparison (i.e. types of burns and mortality-yes/no) were dichotomous. Likewise, there were some errors related to using of abbreviations. These have been corrected accordingly.
Any further responses from the reviewers can be found at the end of the article Since we were collecting data from patients' electronic records, there was possibility that the hospital staffs might have made errors during data entry in the records. This could give rise to information bias in our study. To reduce it to minimum, we cross-checked every information with patients' admission files stored in hospital administratin section itself. Next, all the eligible cases were included in the study. So, there was minimal chance of selection bias as well.

Ethical considerations
Ethical approval was taken from the Institutional Review Committee (IRC no. 006-2021), phect-NEPAL. Before conducting the study, permission was taken from the Department of Burns, Plastic and Reconstructive Surgery. The anonymity of patient information was well-maintained and thus patient consent was waived by the ethical review board.

Statistics
Shapiro-Wilk W test showed that our continuous variables were distributed non-normally, so median and inter-quartile range (IQR) were calculated. Among categorical variables, Chi-square/Fisher's exact test was applied to check the association between independent and dependent variables. Frequency and percentages were presented appropriately in tables. The level of significance was taken as p < 0.05, with a 95% confidence interval (CI) considering a 5% standard error. The analysis was run partly using IBM SPSS version 23, and Stata version 15.

Logistic regression
The dependent variable was mortality outcome, while the rest of the other variables affecting mortality were independent variables. For regression analysis, the outcome of interest as mortality event was taken and those who died in course of treatment were labeled as 1, and those who survived were labeled as 0. Initially binary logistic regression was run to see the effect of individual variables in mortality outcome. Only those variables which were found to be associated in Chi-square/Fisher's exact test and significant continuous variables (age, length of hospital stay, and Baux score) were taken for logistic regression analysis. Later, binary logistic regression analysis was performed to check the exact effect of independent variables adjusting to the rest of the variables to check and nullify the confounding effect of different variables evaluated. Unadjusted odds ratio (OR) for binary logistic regression and adjusted OR were presented with 95% CIs. The pseudo R 2 value for binary logistic regression analysis was 0.5367, indicating that our model predicts a similar outcome in about 54% of observations. Multicollinearity across the studied variables was automatically tested by Stata software and no observed variables were omitted across selected variables for binary logistic regression analysis.

Results
A total of 120 pediatric patients with scald burns were admitted during the two-year period. Out of them, 12 were excluded from our study due to their incomplete records. Finally, 108 cases were analyzed. 23 Demographic and clinical characteristics of the patients are shown in Table 1. The median age of the patients was 2.8 (IQR = 1.6-4.0) years. In total, 61 (56.5%) of them were male and 47 (43.5%) were female. The majority of the burn injuries occurred in the kitchen (57.4%) and most of the patients were from the Low mountain region (60.2%) of Nepal.
Overall, 43 (39.8%) had HCB and 65 (60.2%) had ASB. There was no significant difference in age, gender, place of burns, and pre-hospital intravenous fluid use in the two groups of scald patients. Baux score was higher in HCB (17.5, IQR = 14.0-27.0) but the median body surface area of burns was equal (15%) in both the groups. In total, 13 cases (12.0%) had an infection at presentation, which was higher in HCB (53.8%) than in ASB (46.2%). Neither of them were statistically significant.
The most commonly involved body parts were lower extremities (55.6%) followed by upper extremities (49.1%) and hands (40.7%). Of all the lower extremities burns, 36 (60.0%) were from ASB, and 24 (40.0%) were from HCB. None of these were statistically significant.
Tangential excision was performed in 52 (48.1%) patients, of which 34 (65.4%) and 18 (34.6%) belonged to ASB and HCB groups, respectively. Cases requiring the second excision (7.4%) were equally distributed in both groups (50.0%). Only two patients required flap and amputation and they were from the HCB group.
Overall mortality was 16 (14.8%). Out of total deaths, 12 (75%) were from HCB and four (25%) from the ASB group. This was statistically significant (p = 0.002). Requirement of in-hospital blood transfusion was significantly higher (p = 0.030) in HCB (57.1%) than ASB cases (42.9%). The median duration of hospital stay in both groups was nine days (IQR in HCB: 5.0-13.0, ASB: 3.0-14.0). A significant association was seen between the dependent variable mortality and type of scalds (p = 0.002), place where burn injury took place (p = 0.010), and involvement of body parts like back (p = 0.015), buttock (0.001), perineum (p = 0.026) and lower extremities (0.005) ( Table 2).     Table 3). Binary logistic regression showed higher mortality in HCB in comparison to ASB type of scald. Similarly, burns occurring outside the house had a higher association with mortality. Involvement of the back, buttock, perineum, and lower extremities was found to be associated with higher odds of mortality. However, adjusting independent variables and continuous variables (age, Baux score, length of stay) showed the only type of scald, age, and Baux score found to be associated with mortality. Every one-year increment in age has a 29.0% lower odds of occurrence of mortality event (adjusted OR, 0.71; 95% CI, 0.50-0.99, p = 0.042). Likewise, every one-point increment in Baux score was associated with 19% higher odds of mortality (adjusted OR, 1.19; 95% CI, 1.08-1.32; p < 0.001).
A study in India 18 reported the mortality in pediatric scald burns to be 3.1% which is lower compared to our number (14.8%). The plausible explanation for these differences is the difference in the site of study, sample size, and duration of the study. Another reason could be because our center is a national referral center for burns, so it is likely that more complicated and extensive cases are being referred here. Another study in Kashmir, India 19 showed mortality from scalds to be 10.7% which is comparable to our findings. The most common place of burn event in our study was a kitchen (57.4%) which is comparable to a study (64.7%) done by Riedlinger DI et al. Similarly, grafting was done in 41.2% of patients which also corresponds to our study (46.3%). 5 Another study 6 depicted the incidence of accidental immersion and spill burns as 5.4% and 81.4%, respectively, which differs from our findings where immersion (hot cauldron) and spill burns comprise 39.8% and 60.2%, respectively. Similarly, a Turkish study 3 showed that the most frequent cause of burn was scalding from spillage of hot water (59.7%) followed by bath scalding (i.e. immersion injury) accounting for 26% of cases. This is in line with the percentage of spill burns in our case. Likewise, in Japan, immersion burn (59.3%) was reported to be higher than spill burns (40.7%). 16 This could be due to the provision of the bathtub in developed countries like Japan where there is a high chance of children climbing up and falling into bathtubs.
The most commonly involved body parts were lower limbs (55.6%) and upper limbs (49.1%) in our study. This finding contradicts a study by Drago DA et al, 6 where the upper torso (25.3%) and upper limbs (24.1%) were maximally involved. A Japanese study 16 revealed that the most common sites of immersion injury were trunk and legs (80%) followed by arms, and those of spill burns were trunk (91.7%) followed by head/neck and arms. In our case, the most common body parts involved were lower followed by upper extremities. The same study showed the average body  surface area of scald as 11.3% which is slightly lower than ours (15.0%). Immersion-related burns were more likely to be located on the lower half of the body involving buttocks, thighs, legs, and feet in a French study. 17 Another also found out that most scalds occurred on the upper limbs. 20 These discrepancies could be due to different sample sizes and sites of study. The mean total body surface area reported in a study from Arizona 21 was 8.0% which is much lower than ours (15.0%). The difference could be due to our center being a referral center where complicated burn cases are being referred from all over the country. However, the mean length of hospital stay in this study 21 (8 days) is similar to our study (9 days).
In a study from Ontario, 15 Children with spill burns had a shorter average length of hospital stay (10.8 days) compared to those involved in bathtub immersion burns (18.3 days). In our study, the median duration of hospital stay was equal in both the groups of scalds (nine days) though it was not statistically significant. Likewise, in the same study, the mean age in both scald groups was 1.8 years whereas, in our setting, the median age in HCB and ASB were 3.0 and 2.5 years, respectively. A study from Beijing 22 showed that the scald burns most commonly occurred in the kitchen which supports our result.
There are some limitations of this study that need to be mentioned. The number of cases was lower than expected because of the ongoing coronavirus disease 2019 (COVID-19) pandemic. Lack of complete data also excluded a significant portion of cases from the study. The data also fails to analyze the economic status of children which could make them susceptible not only to a certain type of scald burn but also limits the access to first aid and primary care, hence affecting the outcome. Moreover, this is a single-center study over a short period of time so, the results may not be applicable to the whole country. For that purpose, multi-center studies conducted over a longer duration are recommended.

Conclusions
ASB was more common in our setting. The mortality was higher in HCB group. These were more likely to require an in-hospital blood transfusion compared to patients with ASB. There were no other significant differences between these two groups. The risk of mortality was significantly higher in burn events occurring outside, and those involving the back, buttocks, perineum, and lower extremities. So, special focus should be given to these factors during management.

Centre for Injury Prevention and Research Bangladesh, Dhaka, Bangladesh
Thank you for conducting an exciting study in the field of burn prevention.
The study's title is 'Comparison of accidental pediatric scald burns in a tertiary care center: hot cauldron burns versus accidental spill burns'. The study aimed to compare accidental spill and immersion (hot cauldron) scald burns in a tertiary care center and assess morbidity and mortality caused by them.
The objective of the study is less clear. However, in this study, the authors tried to determine the type of scald burn that is more prevalent and attempted to explore the association between the type of scald burn and the severity of morbidity and mortality. However, to answer this kind of research question, authors need to select an analytical study design.
The conclusion section of the article was not much aligned with your study title and objective.
In my understanding, it is better to redesign the study to see the distribution of scald burn and its consequence.

Are sufficient details of methods and analysis provided to allow replication by others? Yes
If applicable, is the statistical analysis and its interpretation appropriate? Yes